Can MRI biomarkers for hearing loss in enlarged vestibular aqueduct be measured reproducibly?

Abstract

Morphological features of an enlarged endolymphatic duct (ED) and sac (ES) are imaging biomarkers for genotype and hearing loss phenotype. We determine which biomarkers can be measured in a reproducible manner, facilitating further clinical prediction studies in enlarged vestibular aqueduct hearing loss. A rater reproducibility study. Three consultant radiologists independently measured previously reported MRI ED & ES biomarkers (ED midpoint width, maximal ED diameter closest to the vestibule, ES length, ES width and presence of ES signal heterogeneity) and presence of incomplete partition Type 2 from 80 ears (T2 weighted axial MRI). Interclass correlation coefficients (ICC) and Gwet's Agreement Coefficients (AC) were generated to give a measure of reproducibility for both continuous and categorical feature measures respectively. ES length, width and sac signal heterogeneity showed adequate reproducibility (ICC 95% confidence intervals 0.77-0.95, Gwet's AC for sac heterogeneity 0.64). When determining ED midpoint width, measurements from multiple raters are required for "good" reliability (ICC 95% CI 0.75-0.89). Agreement on the presence of incomplete partition Type 2 ranged from "moderate" to "substantial". Regarding MR imaging, the opinion of multiple expert raters should be sought when determining the presence of an enlarged ED defined by midpoint width. ED midpoint, ES length, width and signal heterogeneity have adequate reproducibility to be further explored as clinical predictors for audiological phenotype. We report which ED & ES biomarkers are reproducibly measured. Researchers can confidently utilise these specific biomarkers when modelling progressive hearing loss associated with enlarged vestibular aqueduct.